I could be remembering wrong, but I’m pretty sure they say somewhere that to cut up to 0.5" you have to cut 0.25" from both sides, i.e. you cut 0.25", flip the board and cut 0.25"
Still seems like poor design to have only 16 steps for what supposedly has high precision…
Exactly correct. Our system is designed with tolerances around step size, height measurement, lens variation, and many more factors. You can think of it as an error budget. Step size (which introduces rounding error), measurement variation (which, as noted in the specs, is small), lens variation, and so on take away from the budget. Allowable focus variation (which we’ve been discussing, and which is actually quite large) adds back. They balance closely. Then, the Proofgrade settings are configured to provide exactly enough leeway to balance the budget.
If your material is warped by 0.01", there’s a 50/50 chance it will make it better or worse as you follow the warp. (Note that it’s not the different height - it’s the change in height, so as you cut, part of the material’s at a different height than what the autofocuser measured). Should that change in height make it worse, it will fail to cut through.
It is both true that the material is exquisitely sensitive to height variation, and that the range in which material can be effectively “in focus” relative to the lens is quite large.
I don’t understand how that can be true. It seems to me the weakest link is the huge step size you use: 0.7mm. Why not stop on half steps instead of always doing an even number? Why is the hardware limited to half steps when it has a micro-stepping driver chip?
I think they set the power and speed to compensate for where the focus ends up. Let’s say they have a PG material at .121". They pick the focus height that is the closest, which according to dan leads to the material still being in focus. They then set the power and/or speed to just cut through - in effect compensating for loosy-goosiness of the range of focus. Since the power and/or speed is set to just cut through (in order to minimize char and kerf) then the ‘material is exquisitely sensitive to height variation’. And since they compensate through power and/or speed then ‘the range in which material can be effectively “in focus” relative to the lens is quite large’.
Although I’ve been following this thread loosely, I’m unclear if the claimed area of out of focus will appear on the lens side (Print head) of the work, or on the bed side of the workpiece. From what I gather the area is on the lens (Print Head) side?
Yes the top 0.1". If you enter a number between 0 and 0.4" the motor moves to the corresponding position. How accurate that is to the top of the crumb tray I don’t know because it depends on the homing procedure. But focus distances above 0.417 do not move any further. Previously the software limit was 11mm (0.433") but it was changed recently to 0.5" and the shipping details claim the focus range is 0.5".
Done?
The material rests on a 18” x 20” working bed. The printable area is 11” x 19.5”. The
maximum thickness of material with tray installed is 0.5”; the maximum thickness of
material with the tray removed is 2”. The focus range is 0.5”.
All that was “done” was to change the input limit.
If one could flip over the material being laser cut and then cut from the other side, then that might work. I have no idea how difficult or if even possible that is to do. My GF has been sitting in a box for about a month or so. I’ve been busy with work and redoing the shop space on my garage to accommodate a better workflow and isolate my mess from my family’s mess.
If I would have to guess I’d say it has to do with the current limitations of the camera-dependent height measurements. since the GF estimates the height by snapping a picture of the 45degrees laser projection, which is imho an unreliable and inconsistent method, the software sacrifices precision in lieu of repeatability by rounding to the nearest full step. and it doesnt revert to half-steps when not using the height gauge i.e. when entering the material height manually.
Well they claim the height measurement is accurate to 0.1mm although that is hard to believe after seeing the picture that gets sent to the cloud. Perhaps that is also a lie.
Re Focussing height. Potentially the latest units have the ability to fully focus to 0.5".
However GF are not alerting everyone to the change otherwise those with the old heads would be asking for new upgraded head units. 
Will need someone to measure / check the latest head units…
Perhaps but Scott’s golden email said his machine had 0.5" focal range but it doesn’t. So I doubt they have changed the head since then. How they managed to come up short on all three axes I don’t know.
Seriously, it’s already a beast, just increase the overall dimensions of the machine if you have to, but don’t clip the workable area
Why not stop on half steps instead of always doing an even number? Why is the hardware limited to half steps when it has a micro-stepping driver chip?
If the “in focus” range is larger than one step height, then half steps wouldn’t much improve things, other than perhaps giving a half-step model better “error range” in that the middle of focus would be more precisely at the measured height, rounding to a half step instead of a full step, on average giving you more distance in focus in case of warp.
Since “in focus” isn’t well defined, I’m not sure how to proceed with this line of thought, though. Any ideas?
What I think would certainly improve performance would be if they took multiple height measurements across the cut/engrave path, and used that to control focus height. Then any warp would tend to be detected and corrected for. For “nearly flat” material, I’d think a grid of measurements, such as what 3D printers with Z-probes do, would let you model the “flat” surface and correct for it. Since Laser really is 2D, the math is easier than the mesh calculations that 3D printers use, which rotate the object being printed in 3D space to correct for bed alignment, etc. Admittedly off topic, but it’s pretty amusing propping a Delta printer’s print bed at a 20 degree angle and watch it print a cube that’s tilted 20 degrees so that it’s still a cube flat on the print bed. 
This wouldn’t be enough for truly non-flat surfaces (engraving curved laptops, etc.) but it’d help with gentle warping, or material that’s raised up due to some material on the crumb tray. And that’d give some percentage more reliable lasering!
Indeed! We have exactly that planned.
Wont you get visible artefacts if you change the focus in 0.7mm steps mid print?
Plugging numbers into the Rayleigh formula (waist 0.2mm, beam 5mm, lens focal length 50mm) I calculate the spot will increase 45% to 0.29mm at 0.5" when focused at 0.417". That gives a reduction in power density of slightly more than 2. I don’t see how that can be considered still in focus.
I engraved some composite material tiles on Sunday that were 0.51" thick just to see what would happen. I put them on the crumb tray. The engraving was moved about a 1/2 inch or more to the left of where it should have been. From now on, if it’s over 0.475" thick, I will remove the crumb tray. Alignment has never otherwise been an issue on my machine.
Since it only supports 0.5", it’s dangerous to use thicker material because of the air assist nozzle - it might whack into your material. A hundreth of an inch probably wouldn’t be an issue but with variations in surface height you might bump into it inadvertently.
The focus of anything near 0.45" is when I do a very light score to check my placement because I’m not sure whee the focus gremlins come in
Yeah, it’s not worth it to attempt again for multiple reasons, including material spoilage. The tiles were free samples from EcoSupply and are only about 3" square. Very dense material and very flat. On full power at 500 speed, the GF did not cut very deep into the surface.
@scott.wiederhold, I ran a vaguely similar test using anodized aluminum, to find out just how much the beamwidth changes over the focus range. I observed a range of spot size of about 2:1, along with a notable softening of the edges of the spot.